Re: [Bug-gsl] [bug #39473] more efficient algorithm for 3j, 6j, 9j calculations (gsl_sf_coupling_{3j, 6j, 9j}_e

[Alexey A. Illarionov]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi guys,

Here is a complete rewrite of coupling3j, resulting in much better
accuracy. You can see comparison in report.pdf. (If anyone is
interested I can attach scripts that produced the graph). Now the
"reliable" limit moved from j_sum ~ 80 to ~ 130.

specfunc/coupling3j_regge.c is a autogenerated index permutation table
using simple python2+numpy script. I'm not sure about the GSL policy
on these kind of things. Just in case I have also attached this script.

Patch is against 4858.


Patrick Alken wrote:
> Update of bug #39473 (project gsl):
> 
> Summary: more efficient algorithm for 3j,6j,9j calculations => more
> efficient algorithm for 3j,6j,9j calculations 
> (gsl_sf_coupling_{3j,6j,9j}_e
> 
> 
> _______________________________________________________
> 
> Reply to this item at:
> 
> <http://savannah.gnu.org/bugs/?39473>
> 
> _______________________________________________ Message sent via/by
> Savannah http://savannah.gnu.org/
> 
> 

- -- 
С уважением, Алексей Александрович Илларионов.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.20 (GNU/Linux)

iQEcBAEBAgAGBQJSegqCAAoJEEBWYSFzoNKeROkH/iMJ37MABSdFjuvDe1veBqGH
mzKz5qaJyiAl0AItrgTaOf2VkfaZReiX6HsVhZbHazvfkIE2sOUOgSwCl/N/lY1E
qNBAVCjFjU5du7jSPLuoQ0Kt1+lEf/L0rXbnhdwSU5L9vJ0MbjVHChbYfdyzFf31
/XwxjmpG3k5qiITyC3HBEIMmtOkONCyf3q84BtDhVD2HYevcUZtzfdHh87BV1U6O
94ktUEgYdglzUSDtmd+Zm3kj+H0HLURtuG5gKfZ0Vr4KWDFS0KX3Kl1wQrZZu2fg
0UUC4W3vPLOKO0OVxTKn0qxIKRsD/IeGtwQHymN2ZSKGlAOR7Actx1sIJY+HUEw=
=GRZV
-----END PGP SIGNATURE-----

report.pdf
Description: pdf/pdf

gen_regge_c.py
Description: Text Data

coupling3j.patch
Description: Text Data